Embargo violators and presence in tax havens

Having detected the violators in the first part of the empirical analysis, we aim to compare violators with non-violators in terms of exploitation of tax havens. More specifically, we believe having presence in tax havens will simplify the process of concealing proceeds from illicit trades. As a result, this subsection seeks to test the following hypothesis:

H1: Arms companies with tax haven presence are more likely to violate arms embargoes.

In the following table, we provide summary statistics that compare the detected violators to the non-violators.

10 Note, however, that DellaVigna and La Ferrara (2010) investigated different embargoes in a different time period, namely Angola, Ethiopia, Liberia, Rwanda, Sierra Leone, Somalia, Sudan and Yugoslavia in 1990-2005.

Table 7: Summary statistics of violators and non-violators

Violators Non-violators

Mean Standard deviation Mean Standard deviation Tax haven greater for violators than non-violators regardless of the tax haven lists. However, the difference is more significant when we define havens by the grey list. Among the companies, 84% of the violators are headquartered in an OECD country versus 88% for non-violators. The average company identified as a violator has a bigger corporate group, but it has a lower global presence compared to the non-violators. To draw any inferences regarding our hypothesis, we apply the regression model as defined in subsection 3.2.1.

Table 8: Regression results- embargo violators and presence in tax havens

Dependent variable: 1 if violator, 0 otherwise

(1)

The values in parentheses are the robust standard errors. Statistical significance is denoted with *, ** and ***, representing 10%, 5% and 1% significance level, respectively.

Table 8 is divided into two parts and presents our results from running the regression. In the two initial columns, the dummy for tax haven presence is defined by the black list described in section 4.4. Thus, if the company has presence in at least one jurisdiction that recur on all of the tax havens lists by OECD (2000), Tax Justice Network (2007) and IMF (2008), dTax haven

equals one. In the two final columns, the dummy equals one if the company is present in any of the jurisdictions defined as tax havens by both Tax Justice Network (2007) and IMF (2008).

The main purpose of introducing these two different interpretations of tax havens is to examine whether our results are consistent between the definitions.

In column (1), we only include the dummy for presence in tax haven. The coefficient is, as expected, positive. However, it is not significant, and the interpretation is that companies with tax haven presence, according to the black list, are only 1.3% more likely to violate embargoes than those without. When we in column (2) include our control variables, we observe that this coefficient remains non-significant and decreases to -7%. This indicates that, in our sample, companies with presence in any of these jurisdictions are less likely to violate embargoes.

Consequently, our hypothesis is not supported. The interpretation of the Abroad_Percent is that the likelihood of embargo violation decreases marginally when the global presence increases.

More specifically, a one percentage point increase in global presence leads to a 0.127% decrease in the likelihood of violation. Furthermore, companies that are headquartered in OECD countries are 7.1% less likely to violate embargoes than companies that are not, all else equal.

However, the coefficients for the two latter variables are non-significant. For our last variable, a relative increase in the size of the company leads to a statistically significant increase in the likelihood of violation, at a 10% level.

For columns (3) and (4), we extend the list of jurisdictions to include the ones that OECD did not define as tax havens (grey list). In column (3), we observe that companies with presence in any of these jurisdictions are 14.9% more likely to violate embargoes. The difference in likelihood increases to 27.4% when we include all variables in column (4) and the coefficient is significant at a 5% level for both columns. This is in line with our expectations and we can consequently say that our hypothesis is supported, given that we define tax havens according to the grey list. Moreover, an increase in the global presence has a negative impact on the likelihood of violation. This also applies for companies that are located in OECD countries, and both of these results are in line with our findings from the previous paragraph, although still non-significant. Lastly, we observe that a relative increase in firm size has a slightly negative

effect on the likelihood of violation, and the coefficient is no longer significant. However, this is in contrast with our findings from column (2), both regarding firm size and tax haven presence. A possible explanation for this is that there are one or more relatively large companies that we have identified as violators that are present in a tax haven according to the grey list, but not according to the black list.

Our results appear to be inconsistent between the different definitions of tax havens. The overall probability is both higher and statistically significant when we define tax havens according to Tax Justice Network (2007) and IMF (2008) in the last two columns. This result is somewhat unexpected as the black list includes only the jurisdictions that all the organizations have agreed upon. One might assume that the jurisdictions included in the black list are more used for illegitimate purposes, but this might not be the case. As a result, this emphasizes the lack of consensus between the different lists of tax havens. A more consistent observation is that companies with headquarters in OECD countries have, on average, a negative impact on the probability of embargo violation. One can argue that these companies have a higher threshold of committing crimes compared to non-OECD companies due to higher associated costs (e.g.

reputational) and they thereby contribute to a lower proportion of the violators.

In order to test whether our results are robust, we extend our analysis by examining each embargo in isolation. Given our findings in Table 8, we will only define tax havens according to the grey list.

Table 9: Regression results- comparison of embargoes

Dependent variable: 1 if violator, 0 otherwise

Central African

Republic Libya Democratic Republic

of Congo Somalia Sudan

(1) (2) (3) (4) (5)

dTax haven ^{0.049 0.064 0.043 0.054 0.064}

(0.035) (0.047) (0.064) (0.045) (0.064)

Abroad_Percent -0.111^* -0.068 -0.002 0.069 -0.053

(0.066) (0.106) (0.052) (0.056) (0.041)

dOECD -0.022 -0.042 0.038 0.009 -0.075

(0.069) (0.069) (0.026) (0.011) (0.072)

Ln(Size) -0.006 0.006 -0.006 -0.008 -0.0004

(0.015) (0.010) (0.010) (0.009) (0.009)

Constant 0.091 0.056 0.004 -0.006 0.067

(0.084) (0.077) (0.011) (0.013) (0.072)

Observations 108 108 108 108 108

The values in parentheses are the robust standard errors. Statistical significance is denoted with *, ** and

***, representing 10%, 5% and 1% significance level, respectively.

In Table 9, each column represents the unique embargo. Our dependent variable is equal to one if the company has been identified as a violator within the embargo, zero otherwise. Naturally, Yemen is not included as we did not detect any chains of reactions in this country. As we can observe from the table, companies with tax haven presence are on average more likely to violate embargoes. The results are consistent between the different embargoes, but the coefficients are not significant. However, this could indicate that a tax haven presence matter regardless of the conflict. Among the different embargoes, tax haven presence has the highest effect on embargo violation in Libya and Somalia with an increased likelihood of 6.4%. Contrarily, the Democratic Republic of Congo has the lowest with a likelihood of 4.3%. Lastly, we observe that our control variables appear to be consistent across the different embargoes, except for some insignificant deviations.

Based on our findings from Table 8 and Table 9, there is sufficient evidence to assume that companies with tax haven presence are on average more likely to violate embargoes, given that we define tax havens according to the grey list. Consequently, the results imply that our hypothesis is supported.